True advertising effectiveness is the contribution of advertising spend to verified business outcomes: qualified enquiries, sales, revenue and profit. An attractive platform report does not prove that advertising pays back.
Evaluation should begin not with clicks and impressions, but with the connection between spend, correctly measured conversions and actual sales.
Start with data quality
Check whether enquiries and sales are defined consistently across advertising systems, analytics and CRM. Any differences must be explained before reallocating budget.
- Reconcile key GA4 and GTM events.
- Remove duplicate conversions.
- Check that the source is passed to the CRM.
- Separate new sales from repeat enquiries.
Create a short data map covering the advertising platform, website, analytics system, CRM and financial records. For each transition, record which identifier or event connects the stages.
Look beyond cost per lead
A low lead cost can conceal weak enquiry quality. Compare channels by qualified leads, sales, margin and payback period. The exact metrics depend on the business model, but the logic is consistent:
- CPL shows the cost of an enquiry, not its quality.
- CAC shows the cost of acquiring a new customer.
- ROAS compares revenue with advertising spend.
- ROMI measures the contribution of marketing costs to profit.
- LTV estimates customer value across the entire relationship.
No single metric provides the whole picture. A high ROAS may come from repeat sales to existing customers, while a low CPL may reflect a large number of irrelevant enquiries.
Compare consistent periods
Seasonality, promotions, stock availability, price changes, sales-team performance and decision time all affect results. Compare like-for-like periods and record external changes.
For a long sales cycle, do not evaluate a campaign only on sales from the current week. First establish the typical time from initial contact to payment, then choose the analysis window.
Measure channel contribution, not only the last click
A user may first encounter the brand through display advertising, return via Search and submit an enquiry after a direct visit. A last-click model assigns all value to the final source and hides earlier contacts.
Compare several attribution views and support them with experiments: geographic tests, control groups, Brand Lift or temporary changes in activity. Experiments are not always available, but they separate causal effect from simple correlation more effectively.
Record an improvement plan
The analysis should produce a short list of hypotheses, each with an owner, deadline, verification method and expected effect on a business metric.
Prioritise by potential impact, confidence and implementation effort. Fix data loss and clear technical issues first, optimise campaigns next and scale budget only after that.
A short evaluation process
- Agree on the business objective and definition of the target result.
- Check event, source and CRM data quality.
- Calculate metrics through to the sale, not only the lead.
- Account for seasonality, sales cycle and repeat purchases.
- Compare channels and campaigns with one consistent method.
- Form and test priority hypotheses.
Advertising effectiveness is confirmed not by a platform report, but by a consistent data chain from spend to business outcome.

